San Francisco Bay Area
Tech director with 15+ years building and scaling deep AI & ML products at scale of Amazon, Microsoft and Coupang. Led Search, Recommendations, Ads, Gen AI systems and drove $5B+ measurable incremental growth. Established innovation flywheel with 30+ research publications and patents. Strong ownership, impact, executive presence, cross-functional alignment, urgency and fiscal discipline. Trusted with C-level communication. Homepage: https://pgupta.gitlab.io/
Query Understanding, Retrieval, Relevance, Ranking and Gen AI for Search at Coupang. End-to-End: Innovation, Engineering/Scale, Tech Vision, Drive Business Metrics Deep tech with strong operational excellence, business sense, process championship and team scaling.
• Owner of dynamic bidding for advertisers promoting products on Amazon by modulating their bids up or down during auction based on conversion opportunity. Drive Advertiser success quantified by concrete metrics. • Responsible for defining the strategic vision, innovation, metrics and scientific road-map delivering $XB in revenue, communicating with stakeholders and executive leadership. Manage team of scientists in multiple locations.
Manager for multiple teams at Search Ranking and Relevance. Improve ranking, cold-start and product discovery at Amazon Search. Also led a team on Search Quality Evaluation. • Owner for New Product Discovery in Amazon Search solving cold-start and drive discoverability. Drive Seller success and help shoppers discover relevant new products. • Responsible for defining product roadmap, execution and delivery of end-to-end system in Search Ranking. Delivered +$XB incremental GMS via new product discovery. Manage team of engineers and scientists.
Uber tech lead for ranking features and models powering Search at Amazon for e-commerce, Prime Video, Fresh, Kindle etc. • Optimizing federated search index via consolidation and training ranking models on consolidated index. • Ranking models WW (NA, EU and Asia) by solving challenges from higher diversity candidates and leverage query intent signals. This saved $MM in infra cost and generated $X00M incremental GMS. • Vision and architect of low latency online cold-start framework of predictive priors for behavioral features (to proxy for new products) and drive increase in discovery. Jan 2017 to Jun 2028 -- instrumental part of core ML team driving key initiatives in Amazon across multiple function areas (Payments, Selection, Appstore).
ML/AI research; Computer Science
Open Source Contributions to Learning to rank in Xapian Search Engine Library
Query Auto-completion in Bing web search • Designed algorithm for whole-page optimization for query auto-complete block in Bing search system. Carried data analysis/research, formulation, feature engineering, prototyping and production code for A/B test. • Built DL based offensive query classification system that showed 32% higher recall compared to production system.